None of your Business?
Editor's Note: This article is part of a series written by members of the CAS Committee on Professionalism Education (COPE) and the Actuarial Board of Counseling and Discipline (ABCD). The opinions expressed by readers and authors are for discussion purposes only and should not be used to prejudge the disposition of any actual case or modify published professional standards as they may apply in real-life situations.
Peter N. Netwirkle works for AL Insurance, a small company that writes personal auto coverage in the state of Honesty. He is working on a sophisticated neural network model to be used in target marketing. Peter's source data includes policy details, policyholder information, and claim statistics. Peter has carefully reviewed each potential variable and attempted to identify those factors with the most predictive value. Other elements are ignored by the model.
Peter takes a break from the project to attend the CAS's Predictive Modeling Seminar in Chicago. Over lunch, Bob Bigmouth, FCAS, MAAA, who works for a competitor of AL, tells him about an interesting discovery that he made while working on a model to detect fraud. Bob found that if he grouped policyholders who both lived in apartments and bought high-limit policies that the incidence of fraud was very high-much higher than when each variable was considered separately. The surprising part was that he also found a correlation among these same variables (when considered together) and the overall claim rate. Although policy limits were included in his source data, Peter's model did not consider them.
Should Peter go back to his office and add policy limit as variable in his model?
Bob inadvertently provided information that is proprietary to his company. The knowledge that Bob shared is intellectual property of his company and he should not have discussed it with Peter. Peter's model will be used to gain competitive advantage against Bob's company. If Peter applied this information in the development of his model, he would be in breach of Precept 1 of the CAS's Code of Professional Conduct, which states that an actuary shall act honestly and with integrity. As such, Peter is diminishing the reputation of the actuarial profession.
Why not? While Bob may have violated Precept 9 of the CAS Code of Professional Conduct, nothing in the Code prohibits Peter from using this information. This is great information and Bob freely offered it. Peter isn't using any data or other detailed information from Bob's company. In fact, he will still need to figure out how to incorporate the variable into his model. Also, the correlation Bob discovered seems unusual and there is no guarantee Peter's analysis will produce a similar result. It's certainly worth checking, though, and thanks to Bob, Peter can now try to make his model more accurate.